Download Quality of services solution for efficient communication within a distributed urban traffic control system Ana Maria Nicoleta Mocofan, Răzvan Ghiţă, Vicente Ramón Tomás López, Florin Codruţ Nemţanu

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Recursive InterNetwork Architecture (RINA) wikipedia , lookup

RapidIO wikipedia , lookup

Airborne Networking wikipedia , lookup

Network tap wikipedia , lookup

Asynchronous Transfer Mode wikipedia , lookup

Distributed firewall wikipedia , lookup

Cracking of wireless networks wikipedia , lookup

Distributed operating system wikipedia , lookup

Net bias wikipedia , lookup

IEEE 1355 wikipedia , lookup

Deep packet inspection wikipedia , lookup

UniPro protocol stack wikipedia , lookup

Quality of service wikipedia , lookup

Transcript
U.P.B. Sci. Bull., Series C, Vol. 78, Iss. 1, 2016
ISSN 2286-3540
QUALITY OF SERVICES SOLUTION FOR EFFICIENT
COMMUNICATION WITHIN A DISTRIBUTED URBAN
TRAFFIC CONTROL SYSTEM
Ana Maria Nicoleta MOCOFAN1, Răzvan GHIŢĂ2, Vicente Ramón TOMÁS
LÓPEZ3, Florin Codruţ NEMŢANU4
Modern Urban Traffic Control (UTC) systems are based on cooperative
entities, able to produce optimal signalling strategies independently of traffic
control centres. The more information the junction’s entities receive from their
neighbours, the more effective the distributed system is. But not all the information
is critical. For improving the transmission of data and building an efficient QoS
structure, a simplified UTC packet is proposed. This solution can be deployed in any
real UTC distributed system for guaranteeing that critical traffic parameters are
successfully exchanged between its local decision making units in order to ensure a
fully adaptive operation.
Keywords: urban traffic control, distributed system, adaptive operation, traffic
parameter, communication, quality of services
1. Introduction
Urban traffic control (UTC), as the use of traffic signals to control the flow
of traffic on urban surface streets and arterials [1] have gone through various
implementations in time. They have evolved from fixed-time traffic signal plans
to fully adaptive strategies, able to respond to traffic behaviour in real time [2],
[3].
In time, there have been various developments of UTC systems, starting
with fixed times control strategies, unable to react to spontaneous changes in
traffic progress or incidents, to fully adaptive approaches (centralized or
distributed), based on communication with traffic detectors in the field and real
time computation of signalling times [2], [3].
1
PhD Eng., Department of Telematics and Electronics in Transports, University POLITEHNICA
of Bucharest, Romania, e-mail: [email protected]
2
PhD Eng., Department of Telematics and Electronics in Transports, University POLITEHNICA
of Bucharest, Romania, e-mail: [email protected]
3
Prof., Department of Engineering and Computer Sciences, University Jaume I, Castellon de la
Plana, Spain, e-mail: [email protected]
4
Prof., Department of Telematics and Electronics in Transports, Faculty of Transports, University
POLITEHNICA of Bucharest, Romania, e-mail: [email protected]
176
Ana Maria Mocofan, Răzvan Ghiţă, Vicente Tomás López, Florin Nermţanu
For an UTC system to operate in a fully adaptive mode and provide
efficient signalling plans, it is crucial to receive real time traffic data, measured
and processed by the detection technologies in the field. Just like in an urban
traffic context, the communication network of the system can sometimes be
overloaded or even victim of a Denial of Services (DoS) attack that would prevent
legitimate information to be exchanged between junctions (for a distributed
approach) or between a junction and the traffic control centre (for a centralized
approach). In this way, the fully adaptive operation of a distributed UTC system
would be obstructed [4], [5], [6], [7],[8].
Fig. 1. Centralized and distributed UTC systems concept
By implementing a Quality of Services solution, these issues can be
limited because it would ensure that critical information for the system to operate
in a fully adaptive mode will always reach the intended destination. Moreover, it
allows better flexibility and control over the transmission bandwidth.
However, working autonomously or in an adaptive way could also present
problems. There could be errors in the process of traffic monitoring (raw data,
communications…), Thus, in such systems data quality is required. The data
quality of ITS systems and services has been analysed in several groups and
projects. In 2008, the ITS IEEE has defined a standard regarding the data quality
[9]. In Quantis project [10], a methodology for quality assessment and assurance
for traffic data and information services has been developed. In Easyway project
[11], a document focused on the Data Quality Aspects in ITS Framework has been
developed. This document presents a general vision of the data quality concepts,
examples and it includes a set of best practices. Nevertheless, these studies do not
analyse the communication data quality. They are only focused on the ITS
services itself [12].
Authors have conducted various research for highlighting the importance
of effectively managing the network resources shared between different classes of
information by designing QoS solutions, especially for wireless media [13] and
distributed systems [14], [15], [16].
Quality of services solution for efficient communication within a distributed urban traffic (…) 177
The data packet may need to be customized for different QoS solutions,
therefore, just like proposed in the present paper, other authors have defined
specific node architectures and packet structures suitable to their approach [17].
This paper presents a proposal to solve some congestion problems in the
data and information exchange. The solution is based on the classification of the
traffic parameters in order to guarantee that the most important parameters reach
the destination before the rest.
The present paper is structured as follows: next section presents a
proposed structure for the data packet exchanged at the distributed level in the
UTC system. Then, in section 3, a QoS solution is described, aiming to ensure that
critical traffic information is exchanged between the UTC distributed system’s
entities in order to maintain fully adaptive operation and cooperation even in
difficult conditions of congested communication channels. The solution has been
tested in a network simulation environment. Finally, in section 4 the positive
results are presented.
2. UTC efficient packet structure
In order to implement an efficient quality of services solution for a
distributed UTC system, a simplified structure of data packets has to be defined.
The main idea is to separate the data packets according to the carried information,
so that they can easier be classified and prioritized using their level of importance.
An Ethernet frame has a maximum length of 1500 Bytes. Besides the data
portion, there is a certain overhead added at each layer of the TCP/IP model,
therefore an approximation of 1570 Bytes for the maximum UTC packet can be
assumed. The real packet length for different UTC systems can vary and it is
highly probable that it will never reach the considered 1570 Bytes. However, for
the QoS efficiency demonstration, this maximum message value will be used in
order to cover any future developments. Forecasting is an important side of an
efficient QoS solution, studied and highlighted by various authors as well [18].
Following the TCP/IP model, the proposed UTC packet comprises a
header and a data portion. Depending on the equipment and systems
manufacturers and developers, part of the fields of the proposed UTC packet can
be present or not.
178
Ana Maria Mocofan, Răzvan Ghiţă, Vicente Tomás López, Florin Nermţanu
Fig. 2. Proposed structure for an UTC packet
The UTC header is made of 9 Bytes and contains the required information
for identifying the message type and the 2 ends of the transmission (figure 2). Its
fields are:
•
•
•
•
•
•
•
Message code – uses 5 bits to specify the transmitted information type:
o Control information - communication link and equipment state,
synchronization data, etc.
o Useful data for urban traffic control;
o Error messages such as unknown source/destination, incomplete or
corrupt information, retransmission requests, acknowledgements, etc.
City and area IDs – the area ID is useful to avoid unnecessary processing
of UTC packets by equipments that are not part of the same area;
Source and destination IDs (traffic controller or specialized decision
making unit in the field) – it is an 8 bits coded information, enough to
represent a maximum number of 232 signalized junctions within a city;
Sequence number – a sequence stands for the minimum time unit used for
transmitting the data in the system; that is 1 second.
The maximum number of hops to the destination - depending on the
network topology and the permissions on the direct links of any 2 UTC
junctions, this number can be precisely known or assumed within certain
limits in order to allow a greater network flexibility.
Reference junction and phase IDs – are useful for validating the offset of
each junction in a certain area of the distributed system.
To complete the UTC header, 4 bits are preserved for future needs.
Quality of services solution for efficient communication within a distributed urban traffic (…) 179
3. Quality of Services within a distributed UTC system
Once the UTC packet has been defined, it can be used to build the QoS
simulation scenario for demonstrating how such a solution can ensure fully
adaptive operation of a distributed UTC system, even in difficult conditions of
highly saturated communication links.
The efficiency and accuracy of generating signalling plans is increased if
the UTC decision making units receive all the traffic information they can get.
The main idea is to guarantee that the most important data (vital for the
operation of the system in an adaptive way) would reach the intended destination
no matter what.
A successful solution of ensuring the quality of services for an end-to-end
transmission depends on how well the delay and packet loss parameters are
handled [19]. Therefore, a set of techniques for managing the communication
network’s resources is mandatory.
Message types within a distributed UTC system
In previous research, an analysis was conducted using a traffic simulation
model that permitted the testing of different scenarios using real data collected in
the field.
The traffic network used for the demonstration consisted of 5 successive
intersections in the southern area of Bucharest, Romania, included today into a
modern urban traffic management system with adaptive control operation (BTMS
– Bucharest Traffic Management System).
The data used for the simulation scenarios was the traffic flow information
manually gathered (observed) in the field in March 2012, during morning rush
hour 8 am – 9 am (a mean hourly total of 16,365 vehicles present on the selected
road segment).
Different values of the traffic parameters have been applied and the
variability of the control parameters has been studied. Measures of effectiveness
such as delays, stops, fuel consumption and performance index (linear
combination of the first three, and, optionally, excessive queuing) (Minnesota
Department of Transportation, 2009) have also been analysed.
The results of the analysis show that the UTC decision making
components are able to produce improved signalling plans if they have available
as many traffic information they can get from the field detectors.
The chart in figure 3 confirms this last statement. The optimal values were
achieved when all the traffic parameters allowed by the simulation tool were
known and employed at every step by the system’s decision units (the dotted trend
line at the bottom of the chart). The best value of the network’s performance
180
Ana Maria Mocofan, Răzvan Ghiţă, Vicente Tomás López, Florin Nermţanu
index is the smallest one, translated into reduced values for the traffic delays and
stops [20].
Performance Index ($/h)
280
260
240
220
200
180
160
140
80
82
84
86
88
90
92
94
96
98
100
102
104
106
108
110
112
114
116
118
120
Traffic cycle length(s)
Available data: flow
Available data: flow, speed
Available data: flow, speed, classification
Available data: flow, speed, classification, headway
Fig. 3. Performance values for the simulated UTC system depending on the traffic parameters’
availability
The simulations results and conclusions allowed the defining of a
classification scheme for the traffic variables. This ranking is used in this paper as
a determining step for a quality of services ensuring solution within an UTC
distributed system.
Assigning priorities to UTC packets
In order to maintain a high level of quality for the transmissions, priorities
must be set between different types of messages. It is important to establish the
critical data that must be sent to the destination, even with the price of sacrificing
other information [21].
For advanced classification needs within a QoS solution, a traffic
separation model has been developed in the networking field. DiffServ
(Differentiated Services) uses an 8b traffic marking element called DSCP
(Differentiated Services Code Point) that makes a correspondence to a
prioritization scheme used in legacy systems. DiffServ immediately classifies the
information generated within a network and assigns a DSCP identifier to each
packet. It does not affect the active applications and ensures interoperability with
older equipment [22].
Table 1 shows the classification of UTC information together with a TCP
port identifier assigned to each category for the hardware configuration.
Quality of services solution for efficient communication within a distributed urban traffic (…) 181
Table 1
Classification of UTC data and corresponding TCP port
UTC data packet type
Priority
TCP port ID
Signalling plans
Traffic flow
Headway
Vehicle classification
Speed
Turning percentage
Occupancy
System diagnoses
5
4
3
3
2
2
1
1
50000
50001
50002
50003
50004
50005
50006
50007
Defining the QoS policy for a distributed UTC system
The last step in preparing the quality of services implementation is to
create a policy comprising the classes of information delimited before and
establishing the required transmission parameters.
For area coordination, all that needs to be done is to communicate the
processed signalling plans between directly connected neighbours and together
will be able to apply an optimized strategy. However, as stated before, in order to
ensure a higher efficiency and better performance measures, it is highly
recommended to communicate all the other local traffic and control parameters as
well.
By knowing the size of the data packets about to be transmitted over a
communication channel, plus the overhead due to routing protocols and other
network administrative information, allows the estimation of the required
bandwidth for achieving an efficient data transfer.
For a better demonstration of the QoS functionality, some boundaries have
been set for the amount of bandwidth each type of UTC packet is allowed to use
(high for signalling plans and limited for the rest of the traffic parameters), plus
specific actions in case these values are to be exceeded:
•
•
•
•
All the UTC packets, except diagnoses, should be transmitted according to
new priorities;
Some of the non-UTC packets should be transmitted with lower priority
and reserved bandwidth;
The rest of the non-UTC packets plus UTC diagnoses packets should be
directly dropped;
For all the UTC packets except the signalling plans (that should always be
transmitted) and for all non-UTC packets, violating the exceed limit should
drive to packet drop.
182
Ana Maria Mocofan, Răzvan Ghiţă, Vicente Tomás López, Florin Nermţanu
Table 2 summarizes the actions to be taken if the configured bandwidth is
exceeded or violated (the exceed portion is also consumed).
Table 2
Basic actions for the transmission of the UTC packets over a congested channel
Packet
type
Exceed
action
Violate
action
Signalling
plan
transmit
transmit
Traffic
flow
transmit
with
highest
priority
drop
Headway
transmit
with high
priority
Classification
transmit
with high
priority
drop
drop
Speed
transmit
with
medium
priority
drop
Turning
%
transmit
with
medium
priority
drop
Occupancy
Diagnoses
ICMP
HTTP
transmit
with small
priority
drop
drop
drop
drop
drop
drop
drop
Simulation methodology
For simulating an excessively congested communication channel and
configuring the QoS solution for a distributed UTC system, the following
software tools have been employed:
• GNS3 (Graphical Network Simulator 3) – a network testing
environment offering a graphical user interface and command line
configurable equipments [23];
• Wireshark – a data traffic analyzer for monitoring the network’s
performance [24];
• Colasoft Packet Builder 1.0 – a method of injecting data packets into
a modelling environment that has been used for simulating the UTC
information [25];
• Cat Karat Packet Builder 1.51.200 – also a packet generator used for
building ICMP and HTTP data packets, representing the illegitimate
traffic trying to saturate the UTC network [26].
The test has been run for 2 different situations: normal operation and
congested network (denial of services scenario simulation as a limit).
An informatics attack aiming to produce the denial of services (DoS) is an
attempt of turning the network unavailable for its legitimate users, for a specific or
undefined period.
There are various ways of conducting a DoS attack and usually target the
interruption of a client-server connection. The hacker can achieve this by
interposing himself between these 2 entities and saturating the server with false
requests. Even worse than that, if a system has a poor security mechanism, the
attacker can replace the server and send corrupted information towards the clients
[27]. This kind of attack can seriously impact a centralized UTC system.
Quality of services solution for efficient communication within a distributed urban traffic (…) 183
4. Results and discussions
Fig. 4. Wireshark capture for studying the information flow generated in a junction – marking an
UTC packet type „signalling plan” identified by port number 50000
Under normal operation, the packets are marked according to the QoS
policy applied on equipment R1 (placed in one of the simulated junctions, the
source junction), as highlighted in figure 4. The transmission runs smoothly and
no data loss occurs.
For the second and most relevant scenario, malicious packets were
injected into the network on purpose, to show the improvements offered by the
QoS solution in the event of a DoS attack. Therefore, corrupted or oversized
packets of data were sent throughout the simulated network. The simulations have
been run for 5 minutes in order to evaluate the bandwidth consumption and data
traffic behaviour.
Figure 5 shows the statistics regarding the transmitted packets inside the
established bandwidth interval, the exceeding data packets and the over exceeding
(violating) traffic.
The network attack can be easily identified on the chart, as the number of
ICMP and HTTP packets is huge compared to the legitimate information - 18.3
millions malicious packets over 5 minutes, more than enough to saturate the
communication channel and block the UTC data transmissions from R1 to R2.
Nevertheless, the measures deployed for limiting the available bandwidth
for different types of services turned out to be effective. The green horizontal
portion in figure 5 represents the information that will definitely get to its
intended destination. Thus, UTC data do reach junction 2 in a higher proportion
184
Ana Maria Mocofan, Răzvan Ghiţă, Vicente Tomás López, Florin Nermţanu
than other types of packets, even in difficult conditions of congested
communication channel.
Fig. 5. Guaranteed bandwidth and priority of each type of service
Fig. 6. Successful and unsuccessful transmissions on a highly congested simulated channel
Moreover, figure 6 offers the transmitting percentage of UTC data packets
from junction 1 to junction 2. The signalling plans reach their destination with 0%
losses according to their highest priority and transmission actions in place.
Because the other UTC information is not critical for the system to operate in an
adaptive mode, it has been sacrificed in order to ensure the signalling plans are
being transmitted. Their drop percentage is high, up to 93% for diagnoses packets.
For what concerns the illegitimate data traffic (ICMP and HTTP), it was
99.99% dropped.
5. Conclusions
The current paper builds upon the idea of developing a distributed UTC
system, by customizing a QoS solution for improved outcomes related to seamless
adaptive operations through a higher level of fault tolerance and reliability.
Quality of services solution for efficient communication within a distributed urban traffic (…) 185
Each existing UTC system employs its own information management
methods and protocols, with corresponding security measures. The idea presented
in the current paper is the development of a simplified structure for the UTC data
packet that can be used or adapted to existing or conceptual UTC systems. The
model is based on breaking down traffic information into separated packets
according to the traffic parameters inside. In this way, traffic information flowing
between local junctions can be easier classified and prioritised.
Urban traffic and control parameters needed to be classified according to
their contribution to traffic optimization. This ranking scheme served as the
underlying concept of the proposed Quality of Services presented in the paper.
The aim was to guarantee the critical information availability for each junction,
even in excessively loaded communication channel situations. The signalling
plans are the most important data for ensuring cooperation between neighbouring
nodes and for establishing an optimal control strategy at an area level. Therefore,
their availability and integrity has to be ensured.
After testing the QoS implementation scenario in a virtual simulation
environment and studying its operation when the communication bandwidth is
consumed by undesired information, the main conclusion is that it is a viable
solution for achieving a greater control over the transmission link and maintaining
the UTC system’s fully adaptive operation. Not only does it allow a higher
flexibility and control over the available bandwidth, but also ensures that packets
containing the signalling plans will successfully reach their intended destination
even in difficult, saturated conditions.
REFERENCES
[1] McQueen, B. and McQueen, J. (1999). Intelligent transportation systems architectures,
Norwood, MA, USA, Artech House Publishers
[2] KonSULT (2005). The Knowledgebase on Sustainable Urban Land use and Transport, Leeds,
England,
http://www.konsult.leeds.ac.uk/private/level2/instruments/instrument014/l2_014a.htm
[December 2013]
[3] Minea, M., Grafu, F. D. and Surugiu, M. C. (2007). Sisteme inteligente de transport – aplicaţii,
Bucharest, Romania, Matrixrom
[4] Koonce, P. et al, US Department of Transportation, Federal Highway Administration (2008).
Traffic Signal Timing Manual, cap. 5. Washington, SUA
[5] Malek, S., Denney, R. and Halkias, J. A. (1997). Traffic Signal Control Systems. Advanced
Tarnsportation Management Technologies Participant Notebook, Report, Federal Highway
Administration, Washington, D. C., USA, pp. 3.1 – 3.15
[6] Dunn Engineering Asociates in association with Siemens Intelligent Transportation Systems
(2005). Traffic control systems handbook, Washington D.C., USA
[7] Williams, B. (2008). Intelligent Transport Systems Standards (chapter 5). Artech House, Inc.,
London, England
186
Ana Maria Mocofan, Răzvan Ghiţă, Vicente Tomás López, Florin Nermţanu
[8] Transportation Research Board (2010). Adaptive Traffic Control Systems: Domestic and
Foreign State of Practice. A Synthesis of Highway Practice (chapter 5). Washington, D. C.,
USA
[9] ISO/TR 21707:2008. Intelligent transport systems - Integrated transport information,
management and control - Data quality in ITS systems. Website:
http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=34668
[10] Quantis (2008 - 2010) – QUality AssessmeNt and assurance methodology for Traffic data and
Information Services. Website: www.quantis-project.eu [September 2013]
[11] EasyWay project (2007 – 2013). Website: http://www.easyway-its.eu/ [September 2013]
[12] Jin, J., Gubbi, J., Luo, T. and Palaniswami, M. (2012). Network Architecture and QoS Issues
in the Internet of Things for a Smart City, 2012 International Symposium on
Communications and Information Technologies (ISCIT), Koh Samui, Thailand
[13] Dou, C., Chang, Y. H. and Yan, S. Y (2002). Downlink Throughput Analysis in a WCDMA
Network with Mixed QoS Traffic Classes, The 5th International Symposium on Wireless
Personal Multimedia Communications, Honolulu, Hawaii, USA
[14] Zhang, M., Xie, H. and Boukerche, A. (2009). A QoS-Aware Load Balancing Algorithm for
P2P Based Large-Scale Distributed Virtual Environment, IEEE International Workshop on
Haptic Audio visual Environments and Games (HAVE), Milano, Italy
[15] Chu, J. and Lea, C. T. (2007). Optimal Link Weights for Maximizing QoS Traffic, IEEE
International Conference on Communications, 2007 (ICC '07), Glasgow, Scotland
[16] Tragos, E. Z., Bruno, R., Ancillotti, E., Grochla, K. and Siris, V. A. (2010). Automatically
configured, optimised and QoS aware wireless mesh networks, 2010 IEEE 21st
International Symposium on Personal Indoor and Mobile Radio Communications (PIMRC),
Istanbul, Turkey
[17] Attia, E. O., Amin, A. S.Novel and El Henawy, H. M. (2012). IEEE 802.16 Mesh Node
Architecture to Achieve QoS in Coordinated Distributed Mode, 14th International
Conference on Advanced Communication Technology (ICACT), Pyeong Chan, South
Korea
[18] Yang, Y. and Lung, C. H. (2005). QoS Routing – A Case Study of Time-Dependent Routing,
2005 IEEE International Conference on Communications (ICC), Seoul, South Korea
[19] Chakravorty, R., Kar, S. and Farjami, P. (2000). End-to-end internet quality of service (QoS):
an overview of issues, architectures and frameworks. Proceedings of IEEE International
Conference on Industrial Technology (ICIT), Goa, India
[20] Mocofan, A. M. N., Ghiţă, R., Tomás López, V. R., Nemţanu, F. C. (2013). A Comparative
Assessment of Traffic Control Parameters within an UTC Distributed System. Article
published in Journal of Transportation Engineering (ASCE), USA
[21] Szigeti, T. and Hattingh, C. (2004). End-to-End QoS Network Design: Quality of Service in
LANs, WANs and VPNs, Indianapolis: Cisco Press, IN, USA
[22] Cisco Systems Inc., Implementing Quality of Service Policies with DSCP,
http://www.cisco.com/en/US/tech/tk543/tk757/technologies_tech_note09186a00800949f2.s
html [November 2013]
[23] GNS3 official website, http://www.gns3.net/ [December 2013]
[24] Wireshark official website, http://www.wireshark.org/ [December 2013]
[25] Colasoft official website, http://www.colasoft.com/packet_builder/ [December 2013]
[26] Catkarat packet builder, https://sites.google.com/site/catkaratpacketbuilder/ [December 2013]
[27] Cisco Systems, Inc , Cisco Networking Academy (2007). CCNA Exploration Accessing the
WAN - cap. 4